Running Deterministic Inference#

Basic deterministic inference workflow.

This example will demonstrate how to run a simple inference workflow to generate a basic determinstic forecast using one of the built in models of Earth-2 Inference Studio.

In this example you will learn:

How to instantiate a built in prognostic model
Creating a data source and IO object
Running a simple built in workflow
Post-processing results

Set Up#

All workflows inside Earth2Studio require constructed components to be handed to them. In this example, let’s take a look at the most basic: earth2studio.run.deterministic().

def deterministic(
    time: list[str] | list[datetime] | list[np.datetime64],
    nsteps: int,
    prognostic: PrognosticModel,
    data: DataSource,
    io: IOBackend,
    output_coords: CoordSystem = OrderedDict({}),
    device: torch.device | None = None,
) -> IOBackend:
    """Built in deterministic workflow.
    This workflow creates a determinstic inference pipeline to produce a forecast
    prediction using a prognostic model.

    Parameters
    ----------
    time : list[str] | list[datetime] | list[np.datetime64]
        List of string, datetimes or np.datetime64
    nsteps : int
        Number of forecast steps
    prognostic : PrognosticModel
        Prognostic model
    data : DataSource
        Data source
    io : IOBackend
        IO object
    output_coords: CoordSystem, optional
        IO output coordinate system override, by default OrderedDict({})
    device : torch.device, optional
        Device to run inference on, by default None

    Returns
    -------
    IOBackend
        Output IO object
    """

Thus, we need the following:

Prognostic Model: Use the built in FourCastNet Model earth2studio.models.px.FCN.
Datasource: Pull data from the GFS data api earth2studio.data.GFS.
IO Backend: Let’s save the outputs into a Zarr store earth2studio.io.ZarrBackend.

import os

os.makedirs("outputs", exist_ok=True)
from dotenv import load_dotenv

load_dotenv()  # TODO: make common example prep function

from earth2studio.data import GFS
from earth2studio.io import ZarrBackend
from earth2studio.models.px import FCN

# Load the default model package which downloads the check point from NGC
package = FCN.load_default_package()
model = FCN.load_model(package)

# Create the data source
data = GFS()

# Create the IO handler, store in memory
io = ZarrBackend()

/usr/local/lib/python3.10/dist-packages/modulus/models/module.py:360: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  model_dict = torch.load(

Execute the Workflow#

With all components initialized, running the workflow is a single line of Python code. Workflow will return the provided IO object back to the user, which can be used to then post process. Some have additional APIs that can be handy for post-processing or saving to file. Check the API docs for more information.

For the forecast we will predict for two days (these will get executed as a batch) for 20 forecast steps which is 5 days.

import earth2studio.run as run

nsteps = 20
io = run.deterministic(["2024-01-01"], nsteps, model, data, io)

print(io.root.tree())

2025-01-23 04:37:04.907 | INFO     | earth2studio.run:deterministic:75 - Running simple workflow!
2025-01-23 04:37:05.053 | INFO     | earth2studio.run:deterministic:82 - Inference device: cuda
2025-01-23 04:37:05.298 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:209 - Fetching GFS index file: 2024-01-01 00:00:00 lead 0:00:00

Fetching GFS for 2024-01-01 00:00:00:   0%|          | 0/26 [00:00<?, ?it/s]

2025-01-23 04:37:05.302 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: u10m at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:   0%|          | 0/26 [00:00<?, ?it/s]
Fetching GFS for 2024-01-01 00:00:00:   4%|▍         | 1/26 [00:01<00:25,  1.02s/it]

2025-01-23 04:37:06.318 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: v10m at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:   4%|▍         | 1/26 [00:01<00:25,  1.02s/it]

2025-01-23 04:37:06.344 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: t2m at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:   4%|▍         | 1/26 [00:01<00:25,  1.02s/it]

2025-01-23 04:37:06.374 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: sp at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:   4%|▍         | 1/26 [00:01<00:25,  1.02s/it]

2025-01-23 04:37:06.401 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: msl at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:   4%|▍         | 1/26 [00:01<00:25,  1.02s/it]
Fetching GFS for 2024-01-01 00:00:00:  19%|█▉        | 5/26 [00:01<00:03,  5.71it/s]

2025-01-23 04:37:06.430 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: t850 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  19%|█▉        | 5/26 [00:01<00:03,  5.71it/s]

2025-01-23 04:37:06.456 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: u1000 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  19%|█▉        | 5/26 [00:01<00:03,  5.71it/s]

2025-01-23 04:37:06.482 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: v1000 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  19%|█▉        | 5/26 [00:01<00:03,  5.71it/s]

2025-01-23 04:37:06.508 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: z1000 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  19%|█▉        | 5/26 [00:01<00:03,  5.71it/s]
Fetching GFS for 2024-01-01 00:00:00:  35%|███▍      | 9/26 [00:01<00:01, 10.70it/s]

2025-01-23 04:37:06.535 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: u850 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  35%|███▍      | 9/26 [00:01<00:01, 10.70it/s]

2025-01-23 04:37:06.562 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: v850 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  35%|███▍      | 9/26 [00:01<00:01, 10.70it/s]

2025-01-23 04:37:06.588 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: z850 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  35%|███▍      | 9/26 [00:01<00:01, 10.70it/s]

2025-01-23 04:37:06.614 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: u500 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  35%|███▍      | 9/26 [00:01<00:01, 10.70it/s]
Fetching GFS for 2024-01-01 00:00:00:  50%|█████     | 13/26 [00:01<00:00, 15.62it/s]

2025-01-23 04:37:06.641 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: v500 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  50%|█████     | 13/26 [00:01<00:00, 15.62it/s]

2025-01-23 04:37:06.667 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: z500 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  50%|█████     | 13/26 [00:01<00:00, 15.62it/s]

2025-01-23 04:37:06.693 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: t500 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  50%|█████     | 13/26 [00:01<00:00, 15.62it/s]

2025-01-23 04:37:06.720 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: z50 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  50%|█████     | 13/26 [00:01<00:00, 15.62it/s]
Fetching GFS for 2024-01-01 00:00:00:  65%|██████▌   | 17/26 [00:01<00:00, 20.23it/s]

2025-01-23 04:37:06.746 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: r500 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  65%|██████▌   | 17/26 [00:01<00:00, 20.23it/s]

2025-01-23 04:37:06.772 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: r850 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  65%|██████▌   | 17/26 [00:01<00:00, 20.23it/s]

2025-01-23 04:37:06.798 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: tcwv at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  65%|██████▌   | 17/26 [00:01<00:00, 20.23it/s]

2025-01-23 04:37:06.825 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: u100m at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  65%|██████▌   | 17/26 [00:01<00:00, 20.23it/s]
Fetching GFS for 2024-01-01 00:00:00:  81%|████████  | 21/26 [00:01<00:00, 24.26it/s]

2025-01-23 04:37:06.851 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: v100m at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  81%|████████  | 21/26 [00:01<00:00, 24.26it/s]

2025-01-23 04:37:06.878 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: u250 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  81%|████████  | 21/26 [00:01<00:00, 24.26it/s]

2025-01-23 04:37:06.904 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: v250 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  81%|████████  | 21/26 [00:01<00:00, 24.26it/s]

2025-01-23 04:37:06.931 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: z250 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  81%|████████  | 21/26 [00:01<00:00, 24.26it/s]
Fetching GFS for 2024-01-01 00:00:00:  96%|█████████▌| 25/26 [00:01<00:00, 27.54it/s]

2025-01-23 04:37:06.958 | DEBUG    | earth2studio.data.gfs:_fetch_gfs_dataarray:255 - Fetching GFS grib file for variable: t250 at 2024-01-01 00:00:00_0:00:00

Fetching GFS for 2024-01-01 00:00:00:  96%|█████████▌| 25/26 [00:01<00:00, 27.54it/s]
Fetching GFS for 2024-01-01 00:00:00: 100%|██████████| 26/26 [00:01<00:00, 15.47it/s]
2025-01-23 04:37:07.092 | SUCCESS  | earth2studio.run:deterministic:106 - Fetched data from GFS
2025-01-23 04:37:07.106 | INFO     | earth2studio.run:deterministic:136 - Inference starting!

Running inference:   0%|          | 0/21 [00:00<?, ?it/s]
Running inference:   5%|▍         | 1/21 [00:01<00:24,  1.21s/it]
Running inference:  10%|▉         | 2/21 [00:03<00:38,  2.03s/it]
Running inference:  14%|█▍        | 3/21 [00:06<00:38,  2.13s/it]
Running inference:  19%|█▉        | 4/21 [00:08<00:37,  2.22s/it]
Running inference:  24%|██▍       | 5/21 [00:10<00:36,  2.29s/it]
Running inference:  29%|██▊       | 6/21 [00:13<00:35,  2.35s/it]
Running inference:  33%|███▎      | 7/21 [00:15<00:33,  2.40s/it]
Running inference:  38%|███▊      | 8/21 [00:18<00:32,  2.46s/it]
Running inference:  43%|████▎     | 9/21 [00:21<00:30,  2.51s/it]
Running inference:  48%|████▊     | 10/21 [00:23<00:28,  2.56s/it]
Running inference:  52%|█████▏    | 11/21 [00:26<00:25,  2.59s/it]
Running inference:  57%|█████▋    | 12/21 [00:29<00:23,  2.65s/it]
Running inference:  62%|██████▏   | 13/21 [00:31<00:21,  2.70s/it]
Running inference:  67%|██████▋   | 14/21 [00:34<00:19,  2.74s/it]
Running inference:  71%|███████▏  | 15/21 [00:37<00:16,  2.78s/it]
Running inference:  76%|███████▌  | 16/21 [00:40<00:13,  2.71s/it]
Running inference:  81%|████████  | 17/21 [00:42<00:10,  2.60s/it]
Running inference:  86%|████████▌ | 18/21 [00:44<00:07,  2.53s/it]
Running inference:  90%|█████████ | 19/21 [00:47<00:05,  2.51s/it]
Running inference:  95%|█████████▌| 20/21 [00:50<00:02,  2.54s/it]
Running inference: 100%|██████████| 21/21 [00:52<00:00,  2.59s/it]
Running inference: 100%|██████████| 21/21 [00:52<00:00,  2.51s/it]
2025-01-23 04:37:59.821 | SUCCESS  | earth2studio.run:deterministic:146 - Inference complete
/
 ├── lat (720,) float64
 ├── lead_time (21,) timedelta64[h]
 ├── lon (1440,) float64
 ├── msl (1, 21, 720, 1440) float32
 ├── r500 (1, 21, 720, 1440) float32
 ├── r850 (1, 21, 720, 1440) float32
 ├── sp (1, 21, 720, 1440) float32
 ├── t250 (1, 21, 720, 1440) float32
 ├── t2m (1, 21, 720, 1440) float32
 ├── t500 (1, 21, 720, 1440) float32
 ├── t850 (1, 21, 720, 1440) float32
 ├── tcwv (1, 21, 720, 1440) float32
 ├── time (1,) datetime64[ns]
 ├── u1000 (1, 21, 720, 1440) float32
 ├── u100m (1, 21, 720, 1440) float32
 ├── u10m (1, 21, 720, 1440) float32
 ├── u250 (1, 21, 720, 1440) float32
 ├── u500 (1, 21, 720, 1440) float32
 ├── u850 (1, 21, 720, 1440) float32
 ├── v1000 (1, 21, 720, 1440) float32
 ├── v100m (1, 21, 720, 1440) float32
 ├── v10m (1, 21, 720, 1440) float32
 ├── v250 (1, 21, 720, 1440) float32
 ├── v500 (1, 21, 720, 1440) float32
 ├── v850 (1, 21, 720, 1440) float32
 ├── z1000 (1, 21, 720, 1440) float32
 ├── z250 (1, 21, 720, 1440) float32
 ├── z50 (1, 21, 720, 1440) float32
 ├── z500 (1, 21, 720, 1440) float32
 └── z850 (1, 21, 720, 1440) float32

Post Processing#

The last step is to post process our results. Cartopy is a great library for plotting fields on projections of a sphere. Here we will just plot the temperature at 2 meters (t2m) 1 day into the forecast.

Notice that the Zarr IO function has additional APIs to interact with the stored data.

import cartopy.crs as ccrs
import matplotlib.pyplot as plt

forecast = "2024-01-01"
variable = "t2m"
step = 4  # lead time = 24 hrs

plt.close("all")
# Create a Robinson projection
projection = ccrs.Robinson()

# Create a figure and axes with the specified projection
fig, ax = plt.subplots(subplot_kw={"projection": projection}, figsize=(10, 6))

# Plot the field using pcolormesh
im = ax.pcolormesh(
    io["lon"][:],
    io["lat"][:],
    io[variable][0, step],
    transform=ccrs.PlateCarree(),
    cmap="Spectral_r",
)

# Set title
ax.set_title(f"{forecast} - Lead time: {6*step}hrs")

# Add coastlines and gridlines
ax.coastlines()
ax.gridlines()
plt.savefig("outputs/01_t2m_prediction.jpg")

Total running time of the script: (1 minutes 18.549 seconds)

Gallery generated by Sphinx-Gallery